DataFrame.apply returns Series

[craustin]

from pandas import DataFrame, DateRange
dr = DateRange('2011-11-01','2011-11-15')
df = DataFrame(index=dr)
df_or_so_youd_think = df.apply(lambda x:x)
print df_or_so_youd_think

This returns an empty Series.

[wesm]

It's kind of unclear what to do here-- should you pass an empty Series to the passed function to see if the function is a reduction or not? I literally just tinkered with this yesterday

[craustin]

I think you'd pass an empty series to the function. What option would you favor?

[wesm]

Seems like a reasonable route. Maybe should raise an exception if it can't figure things out

[craustin]

Yeah, I haven't seen the underlying code. It just seems that if DataFrame.apply is called on an empty DataFrame and the lambda returns a Series when given an empty Series, the result should be an empty DataFrame. If the lambda returns a scalar, it would make sense the result is an empty Series. Unless I misunderstand apply.

I know this is a pretty degenerate case... and we could work around it - but thanks for thinking about it!

[wesm]

fixed this and added tests in the above commit